Longitudinal study on birthweight and the incidence of endometrial cancer

From 1976 to 2004, we followed 71 751 participants of the Nurses' Health Study and identified 676 invasive endometrial cancer cases. Birthweight, assessed in 1992, was not associated with the incidence of endometrial cancer. No effect modification by menopausal status was observed, but statistical power to detect an interaction was limited.

Endometrial cancer is the commonest invasive gynecologic cancer in women in the United States, with nearly 40 000 new cases diagnosed each year (American Cancer Society, 2007). Birthweight and intrauterine exposures have been related to the risk of breast cancer (Michels and Xue, 2006), childhood leukaemia (McLaughlin et al, 2006) and testicular cancer (Michos et al, 2007). However, data are limited regarding the potential influence of early life exposures on endometrial cancer risk. Potential mechanisms for an association between high birthweight and increased breast cancer risk include exposure to elevated maternal pregnancy oestrogen levels (Petridou et al, 1990;Ekbom et al, 1992), and insulin-like growth factor-I levels (Yang and Yu, 2000;Ostlund et al, 2002), both of which also affect the development of endometrial cancer (Ayabe et al, 1997;Adami et al, 2002). Prenatal nutrition may regulate body size later in life by altering the number of adipocytes (Sayer and Cooper, 2005), reprogram metabolism or influence leptin resistance (Phillips et al, 1999;Breier et al, 2001). As obesity is an established risk factor of endometrial cancer for premenopausal and postmenopausal women (Adami et al, 2002), birthweight as a marker of prenatal nutrition may plausibly influence its development.
Using data from 28 years of follow-up of 71 751 women participating in the Nurses' Health Study (NHS), we examined the association between birthweight and incidence of endometrial cancer later in life.

MATERIAL AND METHODS
The NHS was established in 1976, when 121 700 married registered nurses age 30 -55 years replied to a baseline questionnaire and received questionnaires biennially by mail to update information on demographic, anthropometric, and life style factors, and on newly diagnosed disease. For the current analysis, we excluded women with missing birthweight data, prevalent cases of endometrial or other cancers, and women with hysterectomy at baseline. During follow-up, women were censored if they reported recently diagnosed in situ or invasive endometrial cancer, had a hysterectomy, died or were lost to follow-up.
In 1992, we asked participants to report their own birthweight with the following options: o2.500, 2500 -o3182, 3182 -o3863, 3863 -o4545, X4545 g. On each biennial questionnaire, participants were asked whether they had been newly diagnosed with endometrial cancer during the previous 2 years. The National Death Index was also routinely searched for deaths among women who did not respond to the questionnaires. For endometrial cancers reported by women or their next of kin for those who had died, permission was requested to review the relevant medical records. Study physicians reviewed all medical records and pathological reports to confirm their diagnosis. Cases included in this study were invasive epithelial endometrial cancers with stage greater than IA in the FIGO staging system.
At baseline and during follow-up, we inquired about a variety of personal characteristics including reproductive and life style factors, many being risk factors for endometrial cancer. Information on age, age at menarche, age at first birth and height was obtained at baseline in 1976. Other early life exposures including premature birth (2 þ weeks premature) and duration of having been breast-fed were assessed in 1992, and birth order was queried in 2004. Weight at age 18 was assessed in 1980. Somatotype at age 5 and 10 was assessed in 1988 by asking participants to choose from nine diagrams that best depicted their figure outline at each age. Maternal vital status and family history of endometrial cancer was asked in 1996. Other covariates were inquired and updated during follow-up.
The association between birthweight and incidence of endometrial cancer was analysed using a Cox proportional hazards model. Three covariate-adjusted models were pursued. In the first model (model I), we adjusted only for family history of endometrial cancer and other early life exposures including birth order, duration of being breast-fed and premature birth, in addition to age. In the second (model II), we also included other established or potential risk factors for endometrial cancer. In the third (model III), we additionally adjusted for anthropometric factors, including somatotype at ages 5 and 10, BMI at age 18, and current BMI. Though covariate-adjusted models II and III have better goodness of fit due to additional adjustment for potential risk factors of endometrial cancer, these factors are subsequent to birthweight and could conceivably mediate the effect of birthweight on endometrial cancer risk. Therefore, adjusted model I would be the preferred model to estimate the overall effect of birthweight.
As birthweight was not assessed until 1992, we conducted a sensitivity analysis restricting follow-up to 1992 -2004 and compared the results with the primary analysis using the entire follow-up. We also evaluated potential effect modification by menopausal status, anthropometric factors, and other early-life exposures.

DISCUSSION
In two previous prospective cohort studies, both from Sweden, the risk of endometrial cancer has been investigated in relation to birthweight. Based on 112 cases, the incidence of endometrial cancer in women with birthweight of X4000 g was found to be almost half that of women with birthweight of o3000 g (HR ¼ 0.55 (95% CI 0.36 -1.17)) (McCormack et al, 2005). Based on 73, no significant association was found between birthweight and endometrial cancer (HR ¼ 0.65 (95% CI 0.34 -1.24) comparing the incidence in women with birthweight o2500 g to that in women 43000 g) (Lof et al, 2007). With more than triple the cases of all previous studies combined, results from the current study do not suggest an association.
Endometrial cancer involves cancerous growth of the endometrium. Unlike breast epithelium, in which terminal differentiation occurs largely during the first full-term pregnancy, the endometrial lining undergoes repeated division and differentiation throughout the reproductive life of a woman. Though birthweight has been related to a higher intrauterine exposure to oestrogen (Petridou et al, 1990) and IGF-I (Yang and Yu, 2000), the relation of birthweight to the profile of endogenous hormone later in life is less clear. In one study, birthweight was not associated with overall premenopausal sex hormone levels, but was inversely associated with luteal estrone and estrone sulphate (Tworoger et al, 2006). Similarly, birthweight was found to be only weakly associated with serum IGF-I levels in adulthood (Schernhammer et al, 2007).
The statistical power of this study would allow us to detect a modest association, if it exists. Many risk factors for endometrial cancer and other early life exposures were assessed and accounted for in the analysis. Though birthweight was queried 16 years after the start of follow-up, when compared with the analysis restricted to prospective follow-up, results including the entire follow-up of 28 years did not differ appreciably. We have previously compared self-reported birthweight with data derived from birth certificates and found self-reported birthweight to be highly reliable (r ¼ 0.74) (Troy et al, 1996).